Volume 12, Issue 2, Pages 170-184 (February 2019) Bi-directional Selection in Upland Rice Leads to Its Adaptive Differentiation from Lowland Rice in Drought Resistance and Productivity  Hui Xia, Zhi Luo, Jie Xiong, Xiaosong Ma, Qiaojun Lou, Haibin Wei, Jie Qiu, Hua Yang, Guolan Liu, Longjiang Fan, Liang Chen, Lijun Luo  Molecular Plant  Volume 12, Issue 2, Pages 170-184 (February 2019) DOI: 10.1016/j.molp.2018.12.011 Copyright © 2019 The Author Terms and Conditions

Figure 1 Matrix of Pearson's Correlation Coefficients (r) (Left Triangle) and p Values (Right Triangle) among Drought-Resistant and Agronomic Traits. NT, number of tillers; PH, plant height; FLL, flag leaf length; FLW, flag leaf width; RWC, relative water content; RWL, ratio of water loss; 100GW, 100-grain weight; GW, grain weight; HI, harvest index; RF, relative fecundity; R100GW, relative 100-grain weight; RB, relative biomass; RGW, relative grain weight; RDR, ratio of deep rooting; NDR, number of deep roots; NSR, number of shallow roots; NTR, number of total roots; NRP, number of roots per tiller. Molecular Plant 2019 12, 170-184DOI: (10.1016/j.molp.2018.12.011) Copyright © 2019 The Author Terms and Conditions

Figure 2 Upland and Lowland Rice Collected from China, Their Phylogenetic Relationships, and the Decay of Linkage Disequilibrium across the Genome. (A) Geographic distributions of upland rice in China. Numbers in red indicate provinces where equivalent numbers of upland and lowland rice landraces were collected. 1, Hebei; 2, Henan; 3, Jiangsu; 4, Hunan; 5, Guangxi; 6, Guizhou; 7, Yunnan. (B) A phylogenetic tree of upland, lowland, and common wild rice landraces. (C) LD decay across the genome in upland and lowland rice. Molecular Plant 2019 12, 170-184DOI: (10.1016/j.molp.2018.12.011) Copyright © 2019 The Author Terms and Conditions

Figure 3 Genomic Differentiation and Selection Detected between Rice Ecotypes. (A) The genome-wide Manhattan plot of FST estimated in a 200 kb sliding window with 100 kb step size. Bars under the Manhattan plot describe distributions of genes relevant to drought resistance and GDP (growth, development, and productivity). Blue bars indicate windows containing drought resistance genes. Green bars indicate windows containing GDP genes. Red bars indicate windows containing both drought resistance and GDP genes. (B) The genome-wide Manhattan plot of log2(∏upland/∏lowland) estimated in a 200 kb sliding window with 100 kb step size. (C) The genome-wide Manhattan plot of number of SNPs in a 200 kb sliding window with 100 kb step size. (D) Mean FST of drought resistance (DR)- and GDP-relevant and neutral windows. The bar indicates SE. *** indicates significance at p < 0.001 by independent t-test in comparison with neutral windows. (E) Mean FST of windows with different gradients of ∏upland/∏lowland ratios. The bar indicates SE. ***p < 0.001 indicates significance by independent t-tests in comparison with the genomic average. Molecular Plant 2019 12, 170-184DOI: (10.1016/j.molp.2018.12.011) Copyright © 2019 The Author Terms and Conditions

Figure 4 Signs of Selection Detected in Upland and Lowland Rice. (A) Tajima's D estimated in upland (blue) and lowland rice (orange). A region receiving balancing selection in upland or lowland ecotype is labeled with green or red shading. (B) Composite likelihood ratio (CLR) estimated by SweeD in upland rice. (C) CLR estimated by SweeD in lowland rice. Molecular Plant 2019 12, 170-184DOI: (10.1016/j.molp.2018.12.011) Copyright © 2019 The Author Terms and Conditions

Figure 5 QTLs Identified by Genome-Wide Association Analysis. Symbols right by the QTLs indicate candidate genes with known functions. Four drought-resistance-associated QTLs (annotated by red arrows) are specifically discussed in the article. Molecular Plant 2019 12, 170-184DOI: (10.1016/j.molp.2018.12.011) Copyright © 2019 The Author Terms and Conditions

Figure 6 A Drought Resistance Gene (DCA1) in Tight Linkage with Other Agronomic Genes and a Drought Resistance Gene (LAX1) Possessing Pleiotropic Effects. (A) A phylogenetic tree of upland, lowland, and common wild rice based on DCA1 sequences. (B) FST (maximum value 0.734, outside circle), ∏upland/∏lowland ratio (maximum value 30.1, middle circle), and r2 of each gene to DCA1 within the region of chr10: 15400000–17600000. r2 is the maximum coefficient of correlation for SNPs within each gene with SNPs of DCA1. (C) A phylogenetic tree of upland, lowland, and common wild rice based on the LAX1 sequences. (D) Rate of water loss (RWL) at 2 h in matured leaves of transgenic lines and the wild-type (WT). (E) Relative water content (RWC) measured in osmotically-stressed seedlings. XE16 and XE18 are transgenic lines of the activated form of LAX1. XE19 and XE20 are transgenic lines of inactivated forms of LAX1. Bars indicate SE. *p < 0.05, **p < 0.01, and †p < 0.1 indicate significance differences by independent t-tests in comparisons between transgenic lines and WT. Molecular Plant 2019 12, 170-184DOI: (10.1016/j.molp.2018.12.011) Copyright © 2019 The Author Terms and Conditions